steipete/clawdis
48 skills120k installs18M starsGitHub
Install
npx skills add https://github.com/steipete/clawdisSkills in this repo
1SummarizeSummarize is an agent skill that wires your coding assistant to the summarize.sh CLI so you can compress articles, PDFs, podcasts, and YouTube into readable output during solo-builder research and content workflows. It is built for indie hackers who constantly skim launch posts, competitor landing pages, and long videos but still need accurate takeaways before they validate or write. Invoke it when the user asks what's a link about, to summarize a URL or article, to use summarize.sh, or to transcribe a YouTube video without standing up a separate download pipeline. The skill documents quick-start commands, YouTube summary versus transcript behavior, and how to set API keys per provider. It fits Claude Code, Cursor, and similar agents that can run shell commands with network access. Pair it with note-taking or planning skills after you have a summary so decisions stay tied to source material rather than chat memory alone.17.4kinstalls2Weatherweather is a lightweight agent skill for solo builders who want fast, location-based forecasts inside Claude Code–style workflows without wiring a paid weather API. The skill tells the agent to try `web_fetch` against wttr.in with `format=j2`, summarize `current_condition` and the nearest area, and keep responses short for rain and temperature decisions. Because wttr.in can return HTML tuned for browsers, the instructions explicitly bias toward JSON text extraction and a字符 cap suited to default fetch limits. When fetch tooling is missing, the skill documents a curl fallback to the same service. Typical uses include checking whether a launch demo day will be dry, whether a flight hub is snowed in, or what to pack for an indie hacker meetup—tasks that sit in early research rather than in shipped product code. It is a integration-style utility, not a production monitoring system, and it depends on network access to a third-party endpoint.4.8kinstalls3Tmuxtmux is a compact agent skill for solo builders on macOS or Linux who already run long-lived terminals—REPLs, installers, workers—and need the coding agent to participate without starting duplicate processes. It tells the agent when to prefer tmux versus a normal one-shot shell or a non-interactive background job, then documents the exact commands to list targets, capture recent or full pane output, inject text and special keys safely, and sanity-check prompts from the last lines of scrollback. That makes it especially useful when Claude Code or similar tools orchestrate parallel workers in named sessions. Installation is oriented around Homebrew tmux via declared metadata. The skill does not replace SSH or container orchestration; it keeps human-grade interactive CLIs observable and steerable from the agent loop across building features, shipping checks, and operating stuck processes.4.1kinstalls4Goggog is an agent skill that teaches Claude Code and similar tools to operate Google Workspace through the gog CLI rather than ad-hoc REST snippets or fragile UI scripts. After a one-time OAuth setup—pointing at a client_secret.json, adding accounts with the right service flags, and verifying auth list—the agent can search and send mail, work drafts, and reach Calendar, Drive, Contacts, Sheets, and Docs using documented command patterns. It targets solo builders who live in Gmail and Google Docs while shipping agent-powered automations: customer replies, calendar lookups, file pulls, and lightweight spreadsheet updates from the terminal. The skill is phase-specific to integration work during build and day-to-day operate moments, but the commands themselves support grow and support workflows like inbox triage. You need the gog binary installed (brew formula provided in metadata) and comfort storing OAuth credentials locally. It does not replace a full ETL pipeline; it excels at agent-delegated Workspace chores with explicit flags and account selectors.3.8kinstalls5Nano Pdfnano-pdf is a thin agent skill that teaches your coding agent to drive the nano-pdf command-line tool so you can change titles, fix typos, and adjust slide copy on one page of a PDF using plain English. Solo builders reach for it when a deck or one-pager is almost done but needs a last-mile text change without firing up Acrobat or re-exporting from Figma. The skill documents install via uv, the edit subcommand shape, and practical caveats about page numbering. It is phase-specific to build docs workflows: fast document surgery on files already in your repo or outbound folder. Always verify the rendered PDF after the CLI runs because page indexing can vary by version.3.4kinstalls6Openai WhisperOpenAI Whisper is an agent skill for running the open-source Whisper speech-to-text CLI on the builder’s machine. Solo makers use it when they want transcripts for support logs, content repurposing, or voice-driven agent inputs without sending audio to a hosted API or managing API keys. The skill centers on copy-paste command patterns: pick model size, choose output format such as plain text or SRT subtitles, and optionally translate audio to English. It aligns with OpenClaw-style metadata that expects the `whisper` binary, often installed via Homebrew as `openai-whisper`. First runs download weights to the user cache, which matters for disk and offline expectations. It is a thin integration skill—not a full media pipeline—so agents still need valid file paths and sensible model choices for length and accuracy tradeoffs.3.1kinstalls7Mcportermcporter is a CLI-first bridge between your development environment and Model Context Protocol (MCP) servers, whether they speak HTTP or stdio. Solo and indie builders who ship with Claude Code, Cursor, or Codex use it to discover what tools a server exposes, authenticate (including OAuth), invoke `server.tool` calls with structured arguments, and keep configuration in one place instead of one-off curl scripts. The skill documents the full surface area: listing and schema inspection, flexible call syntax, daemon management for persistent servers, and lightweight codegen to CLI wrappers or TypeScript clients. Install the `mcporter` npm package so the `mcporter` binary is on PATH, then follow the quick-start commands in the skill. It fits the Build phase when you are connecting third-party or custom MCP backends to your agent workflow and need repeatable, inspectable tool calls during implementation.2.7kinstalls8PeekabooPeekaboo is an agent skill that teaches your assistant to use the Peekaboo CLI on macOS for real UI automation—not just static screenshots. Solo and indie builders shipping Mac-native tools, internal utilities, or agent-driven QA can capture screens and menu bar regions, enumerate apps and windows, drive input against snapshotted elements, and execute `.peekaboo.json` automation scripts with a shared cache for faster repeats. The skill points to `peekaboo <cmd> --help`, `peekaboo --version`, and optional `polter peekaboo` for fresh builds. It deliberately catalogs CLI capabilities (excluding separate agent/MCP surfaces) so the model picks the right subcommand instead of improvising shell one-liners. Use when you need repeatable macOS UI checks, demo capture, or scripted desktop workflows from Claude Code, Cursor, or similar agents on Apple Silicon or Intel Macs.2.6kinstalls9ObsidianObsidian is an agent skill for working with Obsidian vaults through the official obsidian command-line interface while the desktop app is running. Solo builders who treat notes as the system of record for ideas, specs, and launch checklists install it so agents can search, open tabs, and mutate notes without guessing vault layout. The skill documents version and help checks, multi-vault naming, parameter quoting, and when plain Markdown file edits remain appropriate versus routing through the live app connection. It spans early research dumps, validation write-ups, build-time task lists, and operate-phase runbooks—as long as Obsidian 1.12.7+ has CLI enabled and the binary is on PATH. Expect integration-style command recipes rather than a note-taking methodology; you still own folder structure and plugins. Review vault names on disk via obsidian.json on macOS when agents need the correct vault= target.2.4kinstalls10HimalayaHimalaya is an agent skill that teaches your assistant to drive the Himalaya CLI for real mailbox work over IMAP and SMTP from the shell. Solo and indie builders who live in Claude Code, Cursor, or Codex can list folders, search envelopes, read messages, draft replies, and file mail without opening a GUI client. The skill documents read/search/write/organize command patterns, points to configuration and MML composition references, and stresses storing credentials in password managers rather than chat logs. Install expects the `himalaya` binary (Homebrew formula documented in metadata). It is a procedural integration skill—not an MCP server—optimized for agents that already have shell access. Use it when you want repeatable email ops during support, billing follow-ups, or operational triage instead of ad-hoc copy-paste from a webmail tab.2.4kinstalls11Apple RemindersApple Reminders is a macOS agent skill that drives the remindctl CLI so you list, add, edit, complete, or delete Apple Reminders and lists from the terminal. Solo builders use it when personal to-dos must appear in the stock Reminders app on iPhone and iPad—not when they need OpenClaw timed alerts, calendar events, or team project boards. Install remindctl with Homebrew, grant Reminders access, and verify with remindctl status before automating. The skill encodes when to clarify “remind me” (Reminders vs cron systemEvent) so agents do not schedule the wrong kind of notification. It is a narrow integration skill: one binary, one OS, one Apple data store—ideal for indie operators who live in Terminal but still want mobile-native reminders.2.4kinstalls12Video Framesvideo-frames is an OpenClaw agent skill that wraps ffmpeg to extract a single image from a video file using a bundled frame.sh script under the skill base directory. Solo builders use it when an agent needs visual evidence from screen recordings, marketing clips, or bug repro videos without opening a GUI editor. Invocation supports the default first frame, an explicit timestamp with --time, or a frame index with --index, always requiring --out for the destination path. Documentation steers JPG for quick sharing and PNG when UI pixels must stay sharp. Metadata declares ffmpeg as a required binary with an optional Homebrew install recipe. It is a narrow integration skill—not a full editor pipeline—but it fits agent workflows that need stills for docs, tickets, or multimodal review during Build.2.3kinstalls13Apple NotesApple Notes is a macOS-only agent skill that wraps Antonio Rodríguez’s memo CLI so your agent can manage Apple Notes from the terminal instead of clicking through Notes.app. Solo builders who already use Apple Notes for ideas, checklists, and client snippets can install memo with Homebrew (or pip from source), grant Automation access when macOS prompts, and then drive list, filter-by-folder, fuzzy search, interactive add/edit/delete, folder moves, and HTML/Markdown export with plain shell commands. It is a narrow integration—no cloud API—but that is the point on Apple silicon Macs where Notes is the system of record. Pair it with agents that need to file meeting outcomes, pull a note before coding, or dump a folder for documentation drafts. Intermediate complexity is mostly permission setup; day-to-day commands are beginner-friendly once memo is installed.2.3kinstalls14ClawhubClawhub is an agent skill that teaches your coding agent to operate the ClawHub CLI against the default registry at clawhub.com (or a custom endpoint). Solo and indie builders use it when they want a repeatable workflow to discover skills, install them into ./skills under the current or OpenClaw workspace directory, keep versions aligned via hash-based updates, and optionally publish their own packages after authenticating. The skill documents the full command surface: search, install, list, update (single skill, all skills, forced upgrades, and non-interactive batches), and publish with explicit metadata flags. It fits builders standardizing on OpenClaw-style skill folders who outgrow ad-hoc git clones or manual copies. Intermediate complexity comes from registry overrides, version pinning, and publish auth—not from deep code changes in your product repo.2.2kinstalls15Coding AgentCoding Agent is an OpenClaw skill for solo builders who need parallel coding capacity without blocking the main conversation. It documents when to delegate—background feature builds, large refactors, issue-to-PR loops, and PR-style review workloads—and when to stay local: trivial edits, lookups, and anything tied to an ACP thread. Installation paths cover Claude Code and Codex CLIs via npm, gated by `skills.entries.coding-agent.enabled`. Hard operational rules reduce flaky runs: always background spawn, correct PTY behavior per engine, capture a real notification route before starting, and require workers to signal success or failure through OpenClaw messaging. Treat it as a supervisor pattern for agentic development stacks where you already run Claude Code or Codex on the machine.2.2kinstalls16Session Logssession-logs is a journey-wide agent skill for OpenClaw users who need forensic access to their own prior conversations. When a user asks what we decided last week or references a parent chat, the agent can query `$OPENCLAW_STATE_DIR/agents/<agentId>/sessions/`—sessions.json for keys and IDs, plus `<session-id>.jsonl` for full transcripts—using jq and rg rather than guessing from truncated context. It is procedural knowledge for solo builders running long-lived coding agents: debug handoffs, recover specs, audit what was promised, or compare branches of work across sessions. Install metadata declares jq and ripgrep binaries. It does not replace product analytics; it is personal agent observability. Use whenever historical chat context is missing from the active window but still on disk.2.2kinstalls17HealthcheckHealthcheck is an agent skill for solo builders who run OpenClaw on a personal Mac, local assistant box, or remote server and need a disciplined security review before touching SSH, firewalls, or remote access. The workflow starts by inferring context—OS version, container vs bare metal, how you connect, public vs LAN exposure, gateway status, backups, disk encryption, and auto-updates—then asks once for permission to run read-only diagnostics. It leans on OpenClaw CLI audits and platform-specific listener and firewall commands so you see real exposure instead of guessing. Only after the access path is confirmed does it propose reversible, staged hardening options as numbered choices, never claiming the product manages OS-level controls. It fits indie operators who want checklist-driven host hygiene without breaking tailnet or SSH access during a late-night deploy.2.1kinstalls18TrelloTrello is an OpenClaw agent skill that connects your coding agent to Atlassian Trello through the public REST API. Solo builders use it to inspect boards, drill into lists, read card descriptions, and create cards without leaving the agent session—handy when shipping tickets should mirror what you just implemented or when you want the agent to file follow-ups. Setup is explicit: API key, user token, and jq for parsing JSON responses from curl commands documented in the skill. It is a thin integration layer, not a full PM methodology; you bring the board structure and naming conventions. Works well for indie teams already living in Trello who want scripted or agent-driven updates during build and light operational triage.2.1kinstalls19GithubThe GitHub skill teaches agents to operate GitHub through the GitHub CLI (`gh`) for issues, pull requests, CI and check logs, comments, reviews, releases, repositories, and raw `gh api` calls. It is aimed at solo and indie builders who ship with Claude Code, Cursor, or Codex and want repeatable, scriptable GitHub actions without leaving the agent session. You reach for it when you need to inspect or open PRs, read check failures, triage issues, or post structured comments—while keeping deep code review on dedicated reading tools. The skill emphasizes authentication hygiene and environment quirks (gateway vs operator HOME), and it deliberately defers local version control to `git`. That separation keeps responsibilities clear: local history and pushes with git; hosted collaboration and automation surfaces with gh.2kinstalls20DiscordDiscord is an OpenClaw-style agent skill that teaches models how to operate Discord through the shared message tool with channel set to discord, instead of inventing one-off API calls. Solo and indie builders use it when a product or support loop already lives in Discord and the agent must post status, read recent traffic, react, run polls, pin items, open threads, or ship component-based UI without breaking server conventions. The SKILL.md encodes operational rules: honor action gates in config, prefer explicit IDs, avoid Markdown tables in outbound text, and use components v2 when rich layout is needed. It assumes channels.discord.token is configured and allowed-tools includes message. The skill is procedural knowledge—copy-paste JSON patterns for common actions—so it fits agent sessions in Claude Code, Cursor, Codex, or generic OpenClaw hosts. It does not replace Discord bot hosting or a full SDK; it standardizes how an already-integrated agent should behave on your server.2kinstalls21Model UsageModel usage is an agent skill that turns CodexBar’s on-machine cost logs into readable per-model summaries for Codex or Claude. Solo and indie builders who run heavy agent sessions need local spend visibility without guessing from vendor dashboards alone. Invoke it when you want the current dominant model by cost on the latest day, or a complete all-models rollup, including optional pretty JSON for dashboards or notes. On macOS you can call through the codexbar binary after a brew cask install; elsewhere you can pipe exported CodexBar JSON into the bundled Python script. It fits the Operate phase as lightweight monitoring for AI tooling bills, especially when you are comparing models or auditing a week of Claude Code versus Codex usage from the same bar.2kinstalls22SonoscliSonoscli is an agent skill that documents how to drive the sonos CLI for local Sonos control: discovery, playback, volume, groups, favorites, queues, and SMAPI Spotify search. Solo builders who automate their workspace or demo environments can hand the agent stable command patterns instead of re-deriving SSDP quirks each time. Installation is via Go module github.com/steipete/sonoscli/cmd/sonos@latest, with troubleshooting for common discover errors such as no route to host on UDP multicast. It is not a cloud media platform skill—it assumes speakers on your LAN and optionally Spotify client credentials for Web API search paths. Prism places it on Build → integrations because it is a terminal integration surface for agents, useful when you are scripting home-office audio, alerts, or personal productivity flows alongside product work.2kinstalls23Blogwatcherblogwatcher is an agent skill that drives the Hyaxia blogwatcher CLI to monitor blogs and RSS/Atom feeds for new articles. Solo and indie builders use it when they want a lightweight, scriptable feed tracker instead of a heavy reader app or bespoke scrapers. The skill documents install via Go, everyday commands such as add, blogs, scan, articles, read, read-all, and remove, and points agents at --help for flags. It fits the Idea phase when you are discovering what competitors publish, which frameworks ship updates, or which niches are active. Your agent can run scans on a schedule or on demand and report new URLs for research notes or content ideas. It is not a full content-management or SEO stack—it is terminal-first feed polling with read-state tracking.2kinstalls24Spotify PlayerSpotify Player is a narrow agent skill that teaches your coding assistant to drive Spotify from the terminal using spogo first and spotify_player as backup. It documents Homebrew installation for both clients, Premium account requirements, cookie-based auth import for spogo, and everyday commands for search, play/pause, skip, device selection, likes, and status. Solo builders use it to keep hands on the keyboard—ask the agent to queue focus music or switch output devices without leaving the repo. It is environment integration knowledge, not a SaaS feature or streaming API product. Install one of the binaries, authenticate once, then invoke the skill when you want agent-mediated playback control during long build sessions.2kinstalls25NotionNotion is an agent skill for the build phase that standardizes how coding agents interact with Notion using the official ntn CLI, covering pages, Markdown bodies, data sources, files, comments, search, Workers, and raw API calls. Solo builders use it when specs, runbooks, or customer research already live in Notion and the agent must read or write without brittle copy-paste. Setup covers global npm install, interactive login, or headless tokens for CI-style runs, with API version pinned for reproducible requests. The skill pushes Markdown-first workflows—create and update pages from strings or files—while keeping escape hatches to documented REST shapes when the CLI surface is thin. It expects ntn or curl on PATH and a Notion integration token scoped to the pages you mutate.1.9kinstalls26Openai Whisper ApiOpenai-whisper-api is a lightweight agent skill for solo builders who want OpenAI speech-to-text without building a custom SDK wrapper. It centers on invoking a bundled shell script against `/v1/audio/transcriptions`, defaulting to `gpt-4o-transcribe` and writing a sibling `.txt` transcript. Builders can swap models for cost or diarization, point at a compatible base URL for proxies, and pass language or speaker hints via prompt flags. It suits agents that ingest voice notes, podcast clips, or support call recordings into downstream summarization or content workflows. Complexity stays beginner-friendly because the contract is file-in, transcript-out with standard CLI dependencies already common on developer machines.1.9kinstalls27Voice CallVoice Call is an OpenClaw agent skill that exposes programmatic phone calls through the voice-call plugin. Solo builders adding voice outreach, verification calls, or conversational loops to an agent product configure a provider—Twilio, Telnyx, Plivo, or mock for offline testing—then drive calls from the CLI or the voice_call tool during agent runs. Actions cover initiating a call with a message, continuing the session, speaking to the user mid-call, ending, and polling status by call ID. The skill is narrow and integration-first: it does not replace a full contact center but documents exactly how OpenClaw expects plugin config, from numbers, and agent-initiated call flows to work. Use when your agent stack already runs OpenClaw and you need telephony as a capability rather than a one-off script.1.9kinstalls281password1password is an agent skill that catalogs practical 1Password CLI (`op`) commands for solo builders who already use 1Password and need agents to fetch or inject secrets safely. It walks through authentication, reading vault references (`op://…`), running commands with secret-backed environment variables, and rendering templates with `op inject`—patterns that keep credentials out of repositories and chat logs. The readme also notes platform requirements: supported shells on macOS, Windows, and Linux, subscription plus desktop app for integration, and per-OS steps to enable CLI integration in 1Password settings. Use it when an agent must wire database passwords, OTP fields, or SSH material into local dev, deployment scripts, or generated config files while staying aligned with vault URIs instead of hardcoded values.1.9kinstalls29SlackThe slack skill packages Slack Web API operations as predictable JSON actions for agents using the OpenClaw slack tool. Solo and indie builders install it when a Claude Code or similar agent must participate in team channels—posting deploy notes, reading recent thread context, reacting with :white_check_mark:, or pinning an incident summary—without hand-writing REST calls each session. Documentation emphasizes reusing channelId and message timestamp IDs already in context, which cuts token waste and retry loops. Coverage spans outbound messaging, in-channel edits and deletes, reaction inspection, pin management, member lookup, and workspace emoji discovery. It is not a Slack app scaffolder or OAuth setup guide; you still provision tokens and channel.slack configuration separately. Best paired with agent workflows that need human-in-the-loop visibility during ship, launch coordination, or operate/support handoffs.1.9kinstalls30Sherpa Onnx TtsThe sherpa-onnx-tts skill documents how agents run a local Node CLI that turns text into audio through sherpa-onnx, keeping inference on your machine instead of a hosted TTS API. Solo builders experimenting with voice assistants, accessibility readouts, or offline demos configure SHERPA_ONNX_RUNTIME_DIR and SHERPA_ONNX_MODEL_DIR (or pass explicit flags), optionally narrow model and tokens files, and write output to a chosen path. The script discovers a lone .onnx file in the model folder and tokens.txt when you do not override paths, which speeds up first-run setup. It fits the Build phase when you integrate speech into an agent toolchain alongside other clawdis skills. It is not a model training guide or a production monitoring stack—you still source ONNX voices, validate latency, and handle licensing for redistribution.1.9kinstalls31Things Macthings-mac wires Prism-listed agents to Things 3 on macOS through the things3-cli binary. Solo builders use it when they want inbox triage, today/upcoming views, project and tag inspection, or quick todo capture without leaving the agent session. Read paths query the local Things database; writes go through the URL scheme with dry-run recommended before mutating tasks. Setup expects Go install of github.com/ossianhempel/things3-cli, Darwin OS, and sometimes Full Disk Access when database reads fail. It is an integration skill for OpenClaw-oriented metadata (emoji, install recipe) but applies to any agent that can run the things CLI. It does not replace Things’ UI for complex project planning—it automates repeatable list/search/add/update commands you already run from the terminal.1.9kinstalls32Openhueopenhue teaches coding agents to operate Philips Hue lighting through the OpenHue CLI connected to a Hue Bridge, which is useful for solo builders who automate home offices, stream setups, or hands-free focus modes while coding. The skill documents when to invoke it—turning lights on or off, dimming rooms, setting scenes or movie mode, and tweaking brightness or color temperature—and when to stop, steering away from non-Hue ecosystems, HomeKit, HVAC, or unrelated entertainment gear. Common flows cover listing lights, rooms, and scenes, then applying on/off, brightness percentages, and mirek values for warm-to-cool whites. Installation is oriented toward macOS brew users via openhue/cli/openhue-cli, with openhue binary required at runtime. It is a narrow integration skill: high leverage if you already standardized on Hue, irrelevant if your smart home stack is mixed-vendor without Hue hardware.1.9kinstalls33OracleOracle is an agent skill that drives the Oracle CLI to send a focused prompt and hand-picked repository files to a second model for review, debugging, refactoring, or design help. Solo and indie builders use it when the primary coding agent has gone deep on a change set and they want an independent pass without pasting the whole repo into chat. The workflow emphasizes a minimal truthful file set, mandatory dry-run preview with token reporting, and choosing browser versus API engines depending on whether you need long Pro-style thinking or fast API calls. Output is explicitly advisory: you still validate against code and tests. The skill documents defaults (gpt-5.5-pro browser, gpt-5.5 API), reattach behavior for hung sessions, and npx fallback when oracle is not on PATH—fitting Ship review and late Build debug loops.1.9kinstalls34Skill CreatorSkill-creator is the meta workflow solo builders reach for whenever an agent capability needs a proper SKILL.md instead of a one-off prompt. It encodes how OpenClaw- and Codex-style skills should look: quoted frontmatter with name and description, body loaded only on trigger, and heavy examples relegated to references/ or deterministic helpers under scripts/. You use it to draft a new skill, audit an overweight package, split long docs, or fix broken YAML before shipping to teammates or publishing to a catalog. The hard rules mirror what experienced skill authors already do manually—no extra README inside the skill tree unless it is a task reference, validate metadata after every edit, prefer noun-phrase descriptions that agents can match reliably. Because skills are reusable procedural knowledge, this package pays off across Idea brainstorming notes, Build integrations, and Ship review checklists alike; invoke it the moment a capability deserves a named, repeatable trigger rather than pasted instructions.1.9kinstalls35Goplacesgoplaces is an agent skill centered on steipete’s CLI for the Google Places API (New). Solo builders use it when they need text search, ambiguous place resolution, rich place details, or review text without maintaining a custom HTTP client. Install via Homebrew, set GOOGLE_PLACES_API_KEY, and run searches with filters such as open-now, minimum rating, location bias, and page tokens. Output defaults to human-friendly terminal text; pass --json when your agent or pipeline needs structured data. Price levels map 0–4, and type filtering follows API limits (one --type per request). Optional GOOGLE_PLACES_BASE_URL supports proxies or test endpoints. It fits indie maps, travel helpers, local discovery features, and agent workflows that enrich prompts with real venue data.1.9kinstalls36Imsgimsg is an OpenClaw-oriented agent skill that documents how to drive the imsg CLI on macOS for iMessage and SMS. Solo builders automating personal or support outreach on Apple’s stack install imsg via Homebrew, grant terminal Full Disk Access and Messages automation, then let the agent list chats, pull conversation history, or send after explicit user request. The skill is deliberately narrow: it does not replace OpenClaw’s normal reply routing, bulk campaigns, or group management, and it points to other tools for Telegram, Slack, and similar channels. Intermediate complexity reflects OS permissions and the need to confirm before any send. Use it when the user names iMessage or SMS; skip it when they only want in-thread replies that the host already routes.1.9kinstalls37Gh Issuesgh-issues is a user-invocable OpenClaw skill for turning GitHub issues into fix attempts and pull requests using the GitHub CLI. Solo maintainers and indie teams can point it at owner/repo or let it infer origin, filter open work, dry-run candidates, then spawn background agents that branch, patch, and open PRs—with optional fork pushes and review-comment processing. Watch and cron modes support semi-autonomous maintenance without babysitting the terminal. The workflow assumes a normal git remote, authenticated gh, and judgment on --yes for unattended runs. It fits operators who already live in GitHub Issues and want agent labor on repetitive fixes, not greenfield feature design. Intermediate complexity reflects git hygiene, tokens, and multi-phase flags spanning triage, implementation, and review response.1.9kinstalls38Gifgrepgifgrep is an agent skill that documents how to use the gifgrep CLI to find GIFs from major providers, browse results in a terminal UI, and materialize assets on disk. Solo and indie builders adopt it when they need fast visual assets for READMEs, launch threads, support macros, or in-chat agent responses without leaving the terminal. The skill encodes install paths (Homebrew tap or `go install`), everyday query patterns, URL-only piping for scripts, and the GIF-Grab chain for review and sharing. It assumes the `gifgrep` binary is on PATH and is aimed at Claude Code, Cursor, and similar agents that can run shell commands safely. It is a task integration, not a planning ritual — invoke when you already know you need a GIF or a still frame, not when you are still validating product scope.1.9kinstalls39Sagsag is a Clawdis/OpenClaw skill that documents how to use the `sag` CLI for ElevenLabs text-to-speech with a familiar macOS `say`-style UX. Indie builders and agent authors install it when they want spoken agent responses, voice previews, or hands-free notifications without building audio pipelines from scratch. The skill covers API key setup, voice listing, model selection, normalization for tricky strings, and model-specific pronunciation tricks—including expressive v3 tags versus SSML breaks on v2 models. It is intentionally narrow: local playback through sag, not a full podcast production suite. Use during Build while stitching voice into automations; skip on hosts where sag is unavailable or when you cannot store an ElevenLabs API key securely.1.9kinstalls40Wacliwacli teaches coding agents to operate the wacli command-line tool for WhatsApp when—and only when—the user clearly wants third-party messaging or to sync and search WhatsApp history. Normal user conversations on WhatsApp through OpenClaw stay on the platform router; this skill prevents agents from grabbing the CLI for everyday chat. Documented flows include QR-based auth, continuous sync, doctor checks, and listing chats with query filters. Safety rules are explicit: confirm recipient and message content before any send, and ask clarifying questions when intent is ambiguous. Solo builders adding agent-driven notifications or CRM-style lookups can treat this as procedural knowledge for a real binary dependency. Install via brew or go per metadata. It is an integration skill tied to external network access and user-owned WhatsApp sessions, so operational discipline matters as much as syntax.1.8kinstalls41GeminiGemini is a thin agent skill for Google’s Gemini CLI, aimed at solo builders who want a second model or Gemma routing without leaving terminal-driven workflows. It shows how to run non-interactive jobs with `-p`, pick models, force structured JSON responses, and append file content from stdin—patterns that fit summarization, codegen side tasks, and quick research passes. The readme also maps extension, skill, hook, and MCP maintenance commands so an orchestrator can align Gemini’s tooling with the rest of an OpenClaw-style stack. Installation expects the `gemini` binary on PATH, commonly via Homebrew. Because authentication is interactive on first launch, the skill reminds agents to complete login before relying on headless runs. Use it across idea research, build-time generation, and ship-time summaries whenever Gemini is the right model—not as a full replacement for your primary coding agent.1.8kinstalls42Bear Notesbear-notes wraps the grizzly CLI so coding agents can manage Bear notes on macOS without leaving the terminal. Solo builders use it to drop session summaries into tagged inboxes, append implementation details to an existing note ID, or pull note content as JSON for downstream planning. Token-gated operations are documented clearly: copy Bear’s API token from Help, store it locally, then use add-text and tag listing when automation needs write access. The skill fits indie workflows where Bear is the system of record for ideas, specs, and ship checklists. It is integration-heavy rather than creative—it expects you already use Bear and grizzly binaries on PATH. Limitations are explicit: Darwin only, Bear must be present, and some callbacks need user-approved tokens.1.8kinstalls43Camsnapcamsnap is an agent skill that wraps the camsnap CLI so solo builders can capture frames and video from RTSP/ONVIF cameras without writing custom ffmpeg scripts. You configure cameras in YAML or with `camsnap add`, then discover devices, grab stills, record short clips, or run motion watches with shell actions. It fits indie operators who want doorbell, office, or lab camera evidence in agent workflows—alerts, incident logs, or quick visual checks before scaling up recording. The skill documents setup, common commands, and a short test capture before longer clips, which reduces failed automation runs. It is a filesystem-and-shell integration, not a hosted video platform; you own credentials, network reachability, and storage on the machine where the agent runs.1.8kinstalls44Songseesongsee is an agent skill that teaches your coding assistant to drive the songsee CLI for audio feature visualization. Solo and indie builders who ship podcasts, music tooling, or ML experiments can generate publication-ready spectrograms and stacked analysis panels without leaving the agent session. The skill covers Homebrew installation of the `songsee` binary, quick starts for full-track and sliced renders, and the full flag surface for visualization types, colormaps, frequency bounds, and PNG/JPG output. It is a narrow integration: the agent learns exact command patterns rather than abstract design theory. Use it when you already have audio on disk or in a pipe and need consistent, scriptable imagery for docs, social clips, or debugging feature extractors. It does not replace a DAW or streaming stack; it complements ffmpeg-capable environments where you want fast, repeatable visuals.1.8kinstalls45Blucliblucli is an agent skill for the BluOS command-line tool `blu`, aimed at solo builders who automate home or studio audio through Bluesound and NAD gear. Instead of memorizing device IDs and XML APIs, you let the agent run discovery, confirm the active target, and issue playback, volume, grouping, and TuneIn commands with the right flags. The skill documents Go-based installation, environment and flag precedence for choosing a player, and JSON output for reliable scripts. It is a tight CLI integration: high leverage if you already run OpenClaw-style agent stacks beside your speakers, niche if you do not own compatible hardware. Prism catalogs it under terminal integrations so you can find it next to other device CLIs. Always confirm the target device before destructive playback changes, especially in shared spaces or live demo setups.1.8kinstalls46Eightctleightctl is an agent skill that teaches your assistant to operate Eight Sleep pods through the eightctl command-line tool, including status, on/off, target temperature, alarms, schedules, audio playback, and base angle. Solo builders use it when they want sleep hardware control from the same terminal session as Claude Code or Cursor, without opening the vendor app for every tweak. Authentication uses ~/.config/eightctl/config.yaml or EIGHTCTL_EMAIL and EIGHTCTL_PASSWORD. The skill stresses that the API is unofficial and rate-limited, so agents should avoid repeated logins and should confirm before changing temperature or alarms. Install is documented as a Go module for the eightctl binary. It is narrowly scoped personal automation—not production infra—but it is a clear CLI integration pattern for agent-driven device ops.1.8kinstalls47Xurlxurl is a CLI wrapper around the X API built for agent-driven sessions where you need to publish, read, search, or engage without handing secrets to the model. Solo and indie builders use it when distribution work should stay in the terminal: drafting a launch thread, replying to mentions, pulling recent search results for positioning research, or uploading media before a post goes live. The skill documents shortcut JSON outputs plus escape hatches to raw v2 endpoints, and it pairs with OpenClaw-style metadata for brew or npm install. It matters because social launch is repetitive and error-prone when tokens leak into chat; xurl encodes a disciplined pattern—check auth status, run commands, never inspect credential files. It is not a content strategy skill; it is the integration layer that makes X actions callable from Claude Code, Cursor, or Codex workflows.1.7kinstalls48BluebubblesBluebubbles is an agent skill for solo builders running OpenClaw who want iMessage as a first-class channel instead of bolting on ad-hoc scripts. It documents how to call the message tool with BlueBubbles actions: outbound text, reactions, message edits, threaded replies, and richer group operations. You collect concrete inputs—targets, message bodies, message GUIDs, and attachment sources—because vague requests like texting a contact still need handles or chat guids. Metadata notes dependency on `channels.bluebubbles` configuration. Use it while building personal agents, family automation, or small-team assistants on macOS-backed BlueBubbles setups. It does not replace App Store launch or SMS compliance work; it gets the integration behavior right inside your agent stack.1.1kinstalls